Evaluation of cluster interconnects for a distributed shared memory

1999 IEEE International Performance, Computing and Communications Conference (Cat. No.99CH36305) Pub Date : 1999-02-10 DOI:10.1109/PCCC.1999.749413

Sumit Roy, V. Chaudhary

{"title":"Evaluation of cluster interconnects for a distributed shared memory","authors":"Sumit Roy, V. Chaudhary","doi":"10.1109/PCCC.1999.749413","DOIUrl":null,"url":null,"abstract":"Clusters of Symmetrical Multiprocessors (SMPs) have recently become popular as low-cost, high-performance computing solutions. The type of interconnection hardware used in these clusters can become a deciding factor in their overall performance. This paper evaluates the performance of three different communication systems, 10 Mbps Ethernet, 100 Mbps FastEthernet and 155 Mbps ATM, using a multithreaded Distributed Shared Memory system, Strings. The raw performance of each network is first measured using netperf. Ten different applications are then used for performance evaluation, including programs from the SPLASH-2 benchmarks, a medical computing application, and some computational kernels. It is found that half of the programs tested are not significantly affected by changes in the bandwidth. Though the ATM network provides the highest overall bandwidth, the remaining applications show that the increase in latency compared to FastEthernet prevents any performance improvement. On the other hand, applications that require only moderately high bandwidths perform substantially better with FastEthernet.","PeriodicalId":211210,"journal":{"name":"1999 IEEE International Performance, Computing and Communications Conference (Cat. No.99CH36305)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"1999-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"1999 IEEE International Performance, Computing and Communications Conference (Cat. No.99CH36305)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/PCCC.1999.749413","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 3

Abstract

Clusters of Symmetrical Multiprocessors (SMPs) have recently become popular as low-cost, high-performance computing solutions. The type of interconnection hardware used in these clusters can become a deciding factor in their overall performance. This paper evaluates the performance of three different communication systems, 10 Mbps Ethernet, 100 Mbps FastEthernet and 155 Mbps ATM, using a multithreaded Distributed Shared Memory system, Strings. The raw performance of each network is first measured using netperf. Ten different applications are then used for performance evaluation, including programs from the SPLASH-2 benchmarks, a medical computing application, and some computational kernels. It is found that half of the programs tested are not significantly affected by changes in the bandwidth. Though the ATM network provides the highest overall bandwidth, the remaining applications show that the increase in latency compared to FastEthernet prevents any performance improvement. On the other hand, applications that require only moderately high bandwidths perform substantially better with FastEthernet.

查看原文本刊更多论文

分布式共享内存的集群互连评价

对称多处理器集群(smp)最近作为低成本、高性能的计算解决方案而流行起来。这些集群中使用的互连硬件类型可能成为影响其整体性能的决定性因素。本文评估了三种不同的通信系统的性能，10 Mbps以太网，100 Mbps快速以太网和155 Mbps ATM，使用多线程分布式共享内存系统，字符串。首先使用netperf测量每个网络的原始性能。然后使用十个不同的应用程序进行性能评估，包括来自SPLASH-2基准测试的程序、一个医疗计算应用程序和一些计算内核。结果发现，一半的测试程序不受带宽变化的显著影响。尽管ATM网络提供了最高的总带宽，但其余的应用程序表明，与FastEthernet相比，延迟的增加阻碍了性能的提高。另一方面，只需要中等高带宽的应用程序使用较快以太网的性能要好得多。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

1999 IEEE International Performance, Computing and Communications Conference (Cat. No.99CH36305)

自引率

0.00%

发文量